· · Kozloff Keith From Sent To Cc Subject Herman Chris@epamail epa gov Wednesday July 16 2003 5 51 PM Keith Kozloff@do treas gov frick walter@epamail epa gov Scott Jeff@epamail epa gov Review of Camisea current spill modelling Keith-This forwards comments by Walter Frick an EPA spill modelling specialist on the MIKE21current spill modelling for Pisco and Paracas Bays done by ESM on 1 behalf of PlusPetrol Earlier I forwarded FYI his detailed comments on the text of the report Rather than work up a more formal paper this forwards his analysis in the form of responses to some policy-relevant propositions I've inserted a bracketed phrase below where ne ded for clarity In summary a threshold problem is that the reliability of the model in the study area is unverified permitting little confidence that the current modelling results accurately describe the effects of a platform or pipeline leak in the bay that documented site-specific conditions could result in sp1ll movement into the bay that verification of the 2-D model is needed as a firs step to identify its strengths and weaknesses as used here given that a 3-D model while much more appropriate could be difficult to develop and run successfully Hope this is helpful regards Please call 564-6463 if questions Best ·Model verification important to demonstrate that the model is reasonably accurate Yes that's a main·goal The 2-d MIKE21 model may be useful for limited purposes however the available data suggests that the model does not seem to capture the variability in water movement in the vicinity of the terminal But with the uncertainties surrounding the Valeport current meter measurements it's hard to tell I am not certain they properly reported low current speed values and the details of the mooring configuration may be introducing variability that is not actually there With inadequate mooring instruments can pump exhibiting forced behavior that ends up affecting the measured signal The model is not verified against existing data i e calibrated or verified to show that the model can predict known conditions Existing conditions appear to show significantly more uniform distribution than the model predicts Right Since they didn't do a complete verification I did some checking myself My figure in the Appendix Fig A l-1 is a limited verification effort Of course my estimates of speed and direction are obtained by zooming the figures in the report and making manual measurements and calculations to determine model velocities at the terminal in the various scenarios They should be able to do it much better Hopefully at least the general approach is valid If it is the point is that the red circles represent model predictions for corresponding representative conditions If the model looked really promising these ___ _c_ir las_N_QU ld a I Lear more in the 2ortions of tq_e their figure with data points in them Instead the predictions are not spread out and many fall in the empty low-speed area Of __ _ course I suspect that they omitted to point out that there would be data points in center if the propeller didn't stick at low current velocities Anyway the results do not conform to my impressi-on of a highly successful model The model is not verified against the floatables data can predict the 'flotables' results 1 i e does not show that the model